Serving DeepSeek-V4.1-Flash at scale on GB300 NVL72, with decoding throughput within 6% of DeepInfer
Serving DeepSeek-V4.1-Flash at Scale: vLLM Matches DeepInfer and Scales to GB300 NVL72
AI BriefingvLLM serves DeepSeek-V4.1-Flash on a single NVIDIA GB300 NVL72 using P/D disaggregation and Wide-EP, reaching 954K tokens/s per GPU and 196 tokens/s for single-user P90 on SemiAnalysis AgentX—7.9× the early single-node result.
Source: vLLM official blog · vllm.ai
Event follow-up1
View all event coverage- DeepSeek fully open-sources its internal reasoning framework, DeepInferWeChat Official Account: DeepSeek · 27 days ago